When it comes to the task of dimensionality reduction, the Principal Component Analysis (PCA) is among the most well known methods. Despite its popularity, PCA is prone to outliers which can be traced back to the fact that this method relies on a covariance matrix. Even with the variety of sophisticated methods to enhance the robustness of the PCA, we provide here in this work-in-progress an approach which is intriguingly simple: the covariance matrix is replaced by a socalled comode matrix. Through this minor modification the experiments show that the reconstruction loss is significantly reduced. In this work we introduce the comode and its relation to the MeanShift algorithm, including its bandwidth parameter, compare it in an experiment against the classic covariance matrix and evaluate the impact of the bandwidth hyperparameter on the reconstruction error.
[1]
Thomas Seidl,et al.
On coMADs and Principal Component Analysis
,
2019,
SISAP.
[2]
Robert C. Bolles,et al.
Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography
,
1981,
CACM.
[3]
Michael J. Black,et al.
Robust Principal Component Analysis for Computer Vision
,
2001,
ICCV.
[4]
Yizong Cheng,et al.
Mean Shift, Mode Seeking, and Clustering
,
1995,
IEEE Trans. Pattern Anal. Mach. Intell..
[5]
Yi Ma,et al.
Robust principal component analysis?
,
2009,
JACM.
[6]
Karl Pearson F.R.S..
LIII. On lines and planes of closest fit to systems of points in space
,
1901
.